Data for firebrands generated from selected structural fuels: Joint Fire Science Program project (15-1-04-4)

In order to have a better understanding of statistical distribution of firebrands' mass, size (projected area), and traveling distance, full-scale firebrand generation experiments were conducted. Full-scale structural components (fence, corner, and roof) and their assemblies were built from typical residential building construction materials. The samples were ignited and exposed to realistic gusty wind traces in a wind tunnel facility in the Insurance Institute for Business & Home Safety (IBHS) Research Center located in Richburg, South Carolina. Water pans were placed downwind to quench the flying firebrands immediately after landing. The distance between the center of the water pans in which the firebrands' landed and the front location of the burning sample was defined as the traveling distance. The firebrands were collected from the water pans and placed in an oven to reach zero percent moisture content. Dried firebrands were scattered on a white sheet. High-resolution pictures were captures of each sheet using a digital camera (Nikon D5600). Following that, an automated image processing algorithm using MATLAB was developed and employed to measure firebrand projected area. Using a digital balance (Sartorius H51, resolution of ±0.0001 gram), firebrand mass was measured. Experiments and raw data collection for this study were conducted from 2016-2017. The result was 50,571 firebrands collected and measured, with 24,149 from structural components and 26,422 from structural assemblies.
The collected firebrands from previous firebrand production experiments using full-scale building components and their assemblies varied between 50 and 1000 firebrands. The sample size of this study is significantly larger than any existing firebrand data sets. This work was based on a statistics-based framework for the sampling and measurement processes in firebrand generation experiments so that the obtained firebrand data can achieve the desired level of statistical reliability. These firebrand data sets are useful in understanding the characteristics and distribution of firebrands generated from various structural fuels. They can be used for developing and training predictive models for the firebrand phenomenon (generation, transport, and ignition), models to predict fire spread in the wildland and wildland-urban interface, and models to estimate risks from wildfire. They are also useful for wildfire mitigation strategies or guidelines to minimize threat and damage from firebrand attacks.
Original metadata was published on 05/20/2020. On 09/01/2021 the data embargo was lifted and the data for this publication became available.